Relationship between MODIS fire hot spot count and burned area in a degraded tropical peat swamp forest in Central Kalimantan, Indonesia

A number of space-borne sensors observe radiant energy at thermal wavelengths. Thermal anomaly data, otherwise known as hotspot data, have been shown to be particularly correlated with the occurrence of active fires (a fire normally with a flaming component and/or smoldering component). Because of a lack of high-quality burned area data, recent studies have used hotspot data as a proxy for burned area when calculating gas emissions or atmospheric pollutants as a result of biomass burning. We argue that the relationship between hotspot data and burned area is spatially variable and strongly dependent on the vegetation type and function. In this article, we explore the relationship between hotspot data and burned area for a region of degraded and partially altered tropical peat swamp forest in southern Kalimantan, Indonesia. MODIS thermal anomaly (MOD14A1) data were used, alongside disaster monitoring constellation (DMC) and Landsat TM data that were used to derive the burnt area, to calculate a figure indicating the average burned area per hotspot (A(R)). Two different levels of hotspot detection confidence are examined in order to ascertain which confidence levels best describe fires in a tropical peat swamp forest environment. Only a relatively small number of fires were detected at the lowest confidence level, but these small number of hotspots improved results. Results for the estimation of burnt area for each hotspot were found to vary between 15 and 16 ha. Omission errors are of the order of 60%; hotspot commission rates are of the order of 8%.